iSixSigma

Surface Finish CPK requirement is killing me!

Six Sigma – iSixSigma Forums Old Forums General Surface Finish CPK requirement is killing me!

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #47649

    ENR
    Participant

    You will have to forgive my ignorance on this, but I am having a hard time understanding how you can achieve a CPK requirement on a feature like surface finish. With only an upper control limit, and in a naturally degrading process (bore grinding), how does one achieve a 1.67 requirement?

    0
    #159051

    Robert Butler
    Participant

      Surface finish measurements are non-normal.  You will have to compute the equivalent Cpk for such values.
      Chapter 8 (Capability for Non-Normal Distributions) of Bothe’s book -Measuring Process Capability.  Has the details. 

    0
    #159068

    Craig
    Participant

    Transforming the data (if non-normal) would give you a more realistic measure of process capability. For some reason, Montgomery uses an inverse transformation in his textbook for this situation (Surface finish). In his example, the transformation made the data appear normal. I didn’t see an actual normality test though.
    Since your data is autocorrelated, is there a way to set up the process such that it degrades slower?
    If not, you would have to change your cutting tool more often if the customer really needs a Cpk of 1.67. (And all else failed!)
    HACL

    0
    #159073

    BadJokeMan
    Participant

    Maybe the best way to improve your Cpk is to make bore grinding more exciting…
     
    … sorry.  That was bad.

    0
    #159077

    Craig
    Participant

    That was corny and yet funny at the same time!

    0
    #159079

    Adam Bowden
    Participant

    Sounds like an automotive requirement. If you have reliable
    surface measurement gages and have run appropriate Gage R&R
    studies I do not see that you have an issue computing it.
    Considering that your CPK number will be based off the mean,
    upper limit and std dev then either you’ll have to get the mean
    pretty close to center limit or as low as possible or really tighten up
    your Std dev (variation).Measuring/calculating – relatively easy – achieving this may be
    more of a challenge – good luck.Adam

    0
    #159899

    R.JAGATH RAKSHAKAN
    Participant

    Absolutely no problem in maintaining Cpk. You must have only a prudent tool monitoring system and tool change/regrind system in place to take care of the process control.
    X mov range  chart is a preffered method of control charting for the process monitoring. One sided tolerence gives more freedom as you can go as near LSL as possible where as in a two sided tolerence you have to maintain centrality!
    thanks,
     
    R.JAGATH RAKSHAKAN

    0
Viewing 7 posts - 1 through 7 (of 7 total)

The forum ‘General’ is closed to new topics and replies.